How might we transform a folk category — in this case, religion — into an analytical category suitable for cross-cultural research? In addressing that question, this book critically explores various approaches to the problem of conceptualizing religion for scholarly purposes, particularly with respect to certain disciplinary interests of anthropologists. The author argues that the most plausible analytical strategy can be based on the idea of family resemblances, especially as that idea has been used and developed in contemporary prototype theory. In the solution proposed, religion is conceptualized as an affair of 'more or less' rather than a matter of 'yes or no,' and no sharp line is drawn between religion and non-religion.